To analysis and improve China Fengyun-2(FY-2, spin-stabilized geosynchronous meteorological satellite) on-orbit Image navigation(IN) performance, an automated landmark matching algorithm(FY-2 automated landmark matching, FALM) for the Visible and infrared spin scan radiometer(VISSR) visible channel has been developed at China National satellite meteorological center(NSMC). FALM is based on a correlation algorithm used to match the observed landmark to the corresponding landmark extracted from the template. FALM can overcome the previous methods’ shortcomings, such as the dependency on ten-year long satellite data and US Defense Mapping Agency data. Each step of FALM, generation of the landmark templates, binarization processing for observed images, image matching between observed images and landmark templates are described. Exclusion of false matching is done by several strict quality measures including cloud contamination detection, prior knowledge check,neighborhood filter and Hypothesis test. 400 days of FY-2data have been processed by FALM and the results have showed that mainly five factors which can influence the FY-2 on-orbit IN performance: orbit control, the integrity of the known IN parameters, the satellite viewing zone adjustment, beta angle computation and the moment of sunshine pressure. Because of FALM’s high processing speed and accuracy, it is ready to put into operation for the FY-2 IN improvement, as well as for operational monitoring purposes, and will be further developed for FY-4. To analyze and improve China Fengyun-2 (FY-2, spin-stabilized geosynchronous meteorological satellite) on-orbit Image navigation (IN) performance, an automated landmark matching algorithm (FY-2 automated landmark matching, FALM) for the Visible and infrared Spread scan radiometer (VISSR) visible channel has been developed at China National satellite meteorological center (NSMC). FALM is based on a correlation algorithm used to match the observed landmark to the corresponding landmark extracted from the template. FALM can overcome the previous methods’ shortcomings, such as the dependency on ten-year long satellite data and US Defense Mapping Agency data. Each step of FALM, generation of the landmark templates, binarization processing for observed images, image matching between observed images and landmark templates are described. Exclusion of false matching is done by several strict quality measures including cloud contamination detection, prior knowledge check, neighborhood filter and Hypoth esis test. 400 days of FY-2data have been processed by FALM and the results have showed showed that mainly five factors which can influence the FY-2 on-orbit IN performance: orbit control, the integrity of the known IN parameters, the satellite viewing zone adjustment, beta angle computation and the moment of sunshine pressure. Because of FALM’s high processing speed and accuracy, it is ready to put into operation for the FY-2 IN improvement, as well as for operational monitoring purposes, and will be developed further for FY-4.